The Slow Decline of Speedy Outfielders

Over the weekend, I wrote a piece for ESPN Insider and FanGraphs+ based around the question of how players like Jacoby Ellsbury have aged previously. There’s a belief among some that speed-and-defense players like Ellsbury are bad bets after they turn 30, since a large chunk of their value is tied to what they can do with their legs, and speed peaks earlier than other skills. However, there’s also data that shows that faster players actually age better than most other player types. Instead of just trying to show you what the aging curves say, though, I figured showing how similar players to Ellsbury actually did might be more appealing.

So, here’s the basic gist of how I went about finding Ellsbury-like players, though I’ll note that the process here is slightly different from the table I used in the ESPN piece, since I have a little more room to explain my thought process and findings here. I went to the leaderboards and set the date range to cover the last 30 years. I set the age filter to cover ages 27 to 29, the same ages as the last three seasons of Ellsbury’s career. To narrow it down to Ellsbury-type players, I used the positional tabs to select only outfielders, and then put a filter in place to cap Isolated Slugging at .180, which gets rid of the power hitters who are not really anything like Ellsbury to begin with. I also put in a minimum of 1,500 plate appearances, so that we only got players who were roughly full time players over those three seasons.

71 different outfielders have met those qualifications over the last 30 years. Of that 71, I wanted to isolate players who had been above average both on offense and on defense, since Ellsbury has contributed value in both aspects of the game and isn’t that comparable to guys who have specialized in one aspect of the game or the other, and had been above average players overall during their age-27 to 29 seasons. So, I filtered those 71 down to players who had accumulated at least +10 WAR over those three years, and had both posted offensive runs above average of 0 or better and defensive runs above average of -10 or better, since defensive metrics aren’t as precise. This narrowed the field down to 17 players, Ellsbury included.

Five of those 17 are still active — Ellsbury, of course, plus Alex Gordon, Michael Bourn, Carl Crawford, and Shane Victorino — and have yet to reach their mid-30s, so they don’t offer us enough information about long term performance trends, so they were also excluded, bringing us down to 12 players who fit the criteria. From there, I also excluded Kirby Puckett and Shane Mack, because both were mostly bat-first players who were rated as average or worse at both baserunning and fielding, so while they squeaked through the filters, neither one is really Ellsbury-like in terms of overall value.

That leaves 10 players, and their performance over their age-27 to age-29 years is listed below.

Name	PA	BB%	K%	ISO	AVG	OBP	SLG	wOBA	wRC+	BsR	Off	Def	WAR
Rickey Henderson	1788	14%	11%	0.165	0.285	0.387	0.450	0.374	133	28	99	6	17
Jacoby Ellsbury	1691	7%	14%	0.166	0.303	0.356	0.469	0.359	123	19	64	34	16
Ichiro Suzuki	2191	6%	8%	0.111	0.328	0.374	0.440	0.350	118	15	64	22	16
Kenny Lofton	1788	9%	11%	0.150	0.324	0.381	0.474	0.372	118	18	60	31	15
Tim Raines	1733	14%	8%	0.163	0.297	0.395	0.461	0.371	135	15	86	-8	14
Andy Van Slyke	1761	10%	18%	0.180	0.271	0.341	0.451	0.352	126	5	56	18	14
Devon White	1914	8%	20%	0.149	0.253	0.314	0.402	0.322	98	10	7	58	13
Derek Bell	1926	7%	18%	0.165	0.285	0.340	0.450	0.343	111	6	32	23	11
Aaron Rowand	1769	6%	18%	0.170	0.283	0.344	0.453	0.346	105	11	22	29	11
Steve Finley	1688	7%	11%	0.127	0.279	0.331	0.406	0.328	106	6	16	29	10
Marquis Grissom	1850	7%	11%	0.148	0.286	0.337	0.435	0.336	100	7	7	36	10
Average	1,841	9%	13%	0.153	0.289	0.354	0.442	0.349	115	12	45	24	13

I know that table is a little overcrowded with information, but I think it’s worth showing how each player did in at various aspects of the game, and how that compares to Ellsbury’s values during the same point in his career. The average at the bottom does not include Ellsbury’s contributions even though he’s listed in the table for reference.

Anyway, the total averages of those 10 comparable players over their age-27 to age-29 seasons are very similar to what Ellsbury has done over the last three years. Their 9%/13%/.153 BB/K/ISO core numbers are basically a match for Ellsbury’s 7%/14%/.166 marks, though Ellsbury’s numbers are actually a little better because the league strikeout rate is higher now than it was in the past. As a hitter, he posted a .359 wOBA/123 wRC+, compared to .349/115 for the group overall. As a group, they averaged +45 OFF/+25 DEF over those three seasons, while Ellsbury’s at +64/+34. The group averaged +13 WAR, while Ellsbury is at +16 WAR. They’re close, but slightly worse across the board. This is a group of comparables that weren’t quite as good as Ellsbury from age-27 to 29, but had similar skillsets at least.

I assume Ellsbury’s going to land a seven year deal this winter, so for comparison, here’s how these players did from ages 30 to 36, though not all of them stayed in baseball for the full seven years, so in their case its ages 30 to the end of their careers.

Name	G	PA	BB%	K%	ISO	AVG	OBP	SLG	wOBA	wRC+	BsR	Off	Def	WAR
Rickey Henderson	869	3819	18%	11%	0.167	0.287	0.418	0.453	0.393	148	39	253	-10	38
Ichiro Suzuki	1115	5148	6%	10%	0.093	0.332	0.377	0.426	0.348	114	43	136	46	35
Kenny Lofton	940	4260	11%	12%	0.136	0.287	0.368	0.422	0.349	107	17	56	69	26
Tim Raines	834	3651	13%	9%	0.125	0.284	0.376	0.408	0.352	116	21	94	-38	18
Devon White	841	3640	7%	18%	0.163	0.272	0.332	0.435	0.336	100	12	11	48	17
Steve Finley	1041	4474	8%	13%	0.201	0.275	0.338	0.476	0.349	110	4	59	-25	17
Andy Van Slyke	557	2363	10%	15%	0.154	0.280	0.354	0.434	0.350	117	6	48	-8	13
Aaron Rowand	509	1865	5%	22%	0.142	0.253	0.310	0.394	0.309	88	-3	-29	12	4
Marquis Grissom	972	3952	5%	16%	0.149	0.264	0.304	0.413	0.311	84	2	-83	-5	4
Derek Bell	314	1373	10%	21%	0.135	0.241	0.323	0.376	0.311	82	0	-33	-17	0
Average	799	3,455	9%	15%	0.147	0.278	0.350	0.424	0.341	107	14	51	7	17

If you’re looking for mid-career collapses, it’s hard to do better (worse?) than Derek “Operation Shutdown” Bell, who basically turned into a pumpkin overnight and was out of baseball by age-33. His offense and defense both vanished, and he became both useless and disruptive when confronted with his own uselessness. That said, Bell is another kind of sketchy comparison who barely snuck through the filters we’d set up; he was a corner outfielder who had up-and-down defensive ratings, so isolating only the 27-29 seasons make him look better than he was for the rest of his career, since those were his only three good years. He also wasn’t all that fast, and only stole 13 bases in his age-29 season, so including him is a little bit of a stretch; I primarily kept him in the sample to avoid accusations of loading up the comp list with favorable selections.

Bell’s not the only guy who fell apart, as Marquis Grissom became a pretty terrible player after age-29, and Aaron Rowand turned into a scrub as well. But, even with those three included, I think it’s worth looking at the average line put up by the entire group.

From 27 to 29, they hit .289/.354/.442, good for a 115 wRC+; in their decline years, they hit .278/.350/.424, good for a 107 wRC+. These are straight averages, not weighted by plate appearances, so Bell counts just as much in these calculations as Henderson does, and we’re not biasing these numbers up by giving more credit to the guys who kept playing. Even with Bell, Rowand, and Grissom as good players gone bad, the overall average for the group barely declined at all at the plate, even though we’re expanding from a three year window filtered for excellence to a seven year window where aging should be kicking in.

We’d expect a lower performance level just from natural regression, since the selection process left us with players who performed at an above average level; it’s a lot easier to go down than up from where we started. We’d also expect a lower performance level from aging, since we’re moving from a player’s peak to his decline years. But, overall, offensive performance at the plate barely changes, with Henderson getting significantly better, and White and Finley getting a little bit better. In terms of wRC+, these players maintained 93% of their 27-to-29 performance, a pretty remarkable figure. If Ellsbury maintained 93% of his 27-to-29 performance, he’d post a 114 wRC+ over the next seven years, a higher mark than he posted in 2013.

That doesn’t mean that these players didn’t get worse, of course. They did, and they got worse in the areas you would expect them to get worse at; speed and defense. As a group, they averaged +4 BSR/+8 DEF per season from 27 to 29, and that went to +2 BSR/+1 DEF from 30 to 36. They got slower and less effective in the field, though again, it’s worth noting that they didn’t lose all of their baserunning and fielding value.

And, like pretty much any other set of players who got older, they played less. From 27 to 29, they averaged 613 PAs per year, and from 30 to 36, it’s 494 PAs per year, essentially a 20% reduction in playing time. But that includes all the zeros put up by guys who didn’t make it to age-36; if you just look at the seasons for which these players actually were still in the game, the average was 592 plate appearances per year. It’s not so much that they all lost 20% of their playing time over the next seven seasons as it is that Bell, Rowand, and Van Slyke performed poorly enough to include nine zero playing time seasons that drags the average way down. You have to account for the risk that Ellsbury could follow a similar path to those three, but we shouldn’t take the PA number as a per-season projection for the next seven years, given that Ellsbury is much less likely to wash out than a few of the inferior comps.

Overall, these 11 Ellsbury-like players averaged +17 WAR over their 30 to 36 seasons; if you exclude Bell from the calculation (which I think one could make a decent case for doing), it goes up to +19 WAR. Just as there are notable players who collapsed, there’s also Henderson (+38 WAR!), Ichiro +(35 WAR), and Lofton (+26 WAR), who were fantastic players over their first seven seasons of their 30s. And remember, in terms of 27-to-29 performance, Ellsbury was in the group with Henderson, Ichiro, and Lofton. These are three best comparisons to Ellsbury’s last three seasons over the last 30 years, and these three aged exceptionally well.

Defense absolutely does peak early and should be expected to decline fairly substantially for any player heading into his 30s. However, history shows that players who are athletic enough to be valuable baserunners and defenders in their twenties are usually good enough athletes to maintain almost all of their offensive value as they get older. For a player like Ellsbury who isn’t just a defensive specialist, that value shouldn’t be expected to decay at the same rate as his defense. He probably won’t continue to be a +10 center fielder, but even if he declines to an average defender in center field, if he maintains a 110 to 115 wRC+, that’s still going to make him a pretty valuable player; there aren’t that many guys in baseball who can play center field and be above average big league hitters.

Depending on how many plate appearances per season you’d project for him, an average decline according to what these 11 comparable players did would leave him as roughly a +3 WAR player for the next seven years. Ellsbury shouldn’t be expected to be a star for the duration of his next contract, but a sustained +3 WAR performance is nothing to sneeze at. Even at just the group’s average of +17 WAR, at $6M per win, that’s $102 million in value. If you bump him up to +20 WAR (basically, omitting Derek Bell as a viable comparsion), you’re at $120 million. If you put more credence on the fact that his recent performance is right line with what Henderson, Lofton, and Ichiro did in those same years, it’s not that hard to talk yourself into +25 WAR, which would equal about $150 million.

There are risks with any player type. Ellsbury is certainly no guarantee, and one serious knee injury could wreck his value in a hurry. At these prices, teams are betting big on areas where some rough assumptions have to be made. But, I think the performance of Ellsbury-like players should at least lend some comfort to teams considering a big contract for him this winter. This player type has historically aged pretty well, and it’s simply not true that they become useless as soon as their speed goes. Ellsbury won’t be an elite defender and baserunner forever, but there’s value in his bat too, and the total package looks to project as a pretty nice piece for the foreseeable future.